Event Date
Speaker: Cornelia Ilin, Assistant Professor of Practice, School of Information, UC Berkeley
Title: "Ped-BERT: Early Detection of Diseases for Pediatric Care"
Abstract: Early detection of diseases is a topic particularly relevant in pediatrics given the well-documented impact of early-life health conditions on later-life outcomes. Yet, early identification for this age group has so far remained uncharacterized, likely because access to relevant health data is severely limited. Thanks to a confidential data use agreement with the California Department of Health Care Access and Information, we are able to develop Ped-BERT: A state-of-the-art deep learning model that accurately predicts the likelihood of 100+ conditions in a pediatric patient’s next medical visit. We link mother-specific pre- and postnatal period health information to pediatric patient hospital discharge and emergency room visits data. Our data set comprises 764k mother-baby pairs and contains medical diagnosis codes as well as temporal and spatial pediatric patient characteristics, such as age and residency zip code at the time of visit. Following the popular bidirectional encoder representations from the transformers (BERT) approach, we pretrain Ped-BERT via the masked language modeling objective to learn embeddings for the diagnosis codes contained in our data. We then continue to fine-tune our model to accurately predict diagnosis outcomes for a pediatric patient's next visit, given the history of previous visits and, optionally, the mother's pre- and postnatal health information. We achieve an area under the receiver operator curve [AUROC] 0.911 and an average precision score [APS] of 0.363. Further, we evaluate Ped-BERT for fairness by assessing whether prediction errors are uniformly distributed across mother-baby demographics and health characteristics subgroups or if specific subgroups are more susceptible to prediction errors than others.
Speaker's webpage (links to UC Berkeley): https://www.ischool.berkeley.edu/people/cornelia-paulik#profile-main
Seminar Date/Time: Thursday October 26, 2023, at 4:10pm
Location: Mathematical Sciences Building 1147 (Colloquium Room)
Refreshments: 3:30pm, MSB Courtyard